Technical Condition Monitoring Methods to Manage the Redundancy of Systems. Part IV: Algebraic, Logical, and Combined Fault Diagnosis Methods

Abstract. Redundancy management in a technical system involves a monitoring procedure to reconfigure the system as needed. The four-part survey presents modern diagnosis methods for dynamic systems as an integral function of monitoring. Part IV of the survey is devoted to promising approaches to integrating (combining) various diagnostic models and methods, according to the authors’ viewpoint. The scheme with a bank of models and a classifier is discussed. Algebraic fault diagnosis methods based on the linear dependence of observation samples using the apparatus of maximum-rank zero matrix divisors are analyzed. The method of parity equations is extended to the multidimensional case. A fault diagnosis approach stemming from the general principle of isomorphism is considered. Logical monitoring methods described include, first, fault diagnosis based on direct and inverse triplex models obtained from Failure Mode and Effects Analysis (FMEA) results and, second, pair monitoring, which reduces dependence on the burdensome assumption of ideal (fault-free) diagnostic tools. General principles of the application of artificial intelligence in fault diagnosis tasks are given.

Keywords: residual, bank of models, linear dependence of samples, matrix canonizers and zero dividers, Failure Mode and Effects Analysis (FMEA), logical network, direct and reverse logic arithmetic, commutative diagram of morphisms, combination of methods, logical pair monitoring, indicator matrix, artificial intelligence systems.